Patient choice modelling: how do patients choose their hospitals?

Area of Science:

• Health Services Research
• Econometrics
• Transportation Geography

Background:

• Predicting future hospital admissions is crucial for healthcare planning.
• Patient choice models help understand healthcare utilization patterns.
• Understanding factors influencing hospital selection is key for resource allocation.

Purpose of the Study:

• To compare the Multinomial Logit and Utility Maximising Nested Logit models in predicting patient hospital choice.
• To identify key factors influencing patient selection of hospitals for both elective and emergency admissions.
• To assess the validity of nested choice structures in patient hospital selection.

Main Methods:

• Fitting Multinomial Logit and Utility Maximising Nested Logit models to real-world hospital admission data.
• Analyzing data from over 200,000 admissions in Derbyshire, UK.
• Considering patient proximity, hospital capacity (beds, parking), and publicly available National Health Service (NHS) data (waiting times, patient scores).

Main Results:

• The Multinomial Logit model was successfully fitted to the data.
• The Utility Maximising Nested Logit model suggested that nesting choices by city or town is likely invalid for this dataset.
• Patient distance to the hospital emerged as a primary determinant of hospital choice, outweighing online statistics.

Conclusions:

• Distance is a more significant factor in patient hospital choice than internet-based statistics.
• The Utility Maximising Nested Logit model's assumptions may not hold for patient hospital selection.
• Multinomial Logit offers a viable approach for modeling patient hospital choice, with distance as a key predictor.